Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.circuit.ai/llms.txt

Use this file to discover all available pages before exploring further.

Circuit has five core concepts that work together. Understanding these will help you navigate the rest of the documentation.

​
Workspaces

A workspace is the top-level container for your organization. Agents, indexes, and groups all belong to a workspace. Users are associated with a workspace through roles, and a single user can belong to multiple workspaces. Workspace administrators have full visibility across the workspace. They can manage users and groups, configure agents, and create and manage indexes.

​
Agents

An agent is an AI assistant configured to answer questions using your indexed documents. Each agent has:
  • A domain that shapes its behavior (field enablement, customer service, RFP, general purpose, and others)
  • A reference scope that defines which indexes it can search
  • A set of tools it can use (search, read documents, fetch web content, and more)
  • Optional skills and actions for specialized workflows
Administrators create and configure agents, then share them with users or groups.

​
Indexes

An index is a container for your workspace’s documents. Think of it as a structured, searchable collection that Circuit can search through. You upload documents to an index: PDFs, Word files, spreadsheets, presentations, CAD drawings, HTML pages, and more. Circuit automatically processes each document using a multi-stage pipeline that extracts content and analyzes structure. This enables agents to find relevant information by meaning, context, and relationships, not only by keywords. You can create multiple indexes to separate content by security boundary or audience. For example, you might keep internal engineering documents in one index and customer-facing materials in another so you can control which users and agents have access to each.
Keep related documents in the same index. An agent can search across multiple indexes, but grouping related content together improves search accuracy.

​
Chat

Chat is how users interact with agents. You type a question or describe what you need, and the agent:
  1. Searches through its configured indexes
  2. Reads relevant document sections
  3. Reasons over the information
  4. Responds with an answer that cites its sources
Chat supports text, images, document references, and multi-turn conversations. You can ask follow-up questions, request clarification, or tell the agent to try a different approach.

​
Feedback

Every chat response has thumbs up and thumbs down buttons. This feedback goes directly to the Circuit team and your administrators, helping them:
  • Identify areas where the agent performs well
  • Spot gaps in the indexed documents
  • Improve agent configuration and data quality over time
Your feedback makes Circuit better for everyone in your workspace.